Signing: FrostSigningManager refuses a wallet statement proposal whose
guardedness the room's own certificate does not allow, before it is
shown, reading the policy off this device's key state.
Reading: GroupWallet.newestAmong takes the policy and drops a statement
it does not allow before rule 1 -- under a guarded policy a plain
statement is no wallet even alone, so an older client's plain wallet has
no addresses derived, no coins watched and no spends signed. walletFor
passes GroupWalletManager.walletPolicyFor. An undecided room reads
exactly as before. GuardianManager.verify refuses a forwarded statement
whose certificate's policy contradicts its keys or decided unguarded.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GroupWalletViewModel reads the room's own wallet policy through
ChatRepository.walletPolicyOf: a parent that decided guarded leaves the
guarded wallet alone, one that decided unguarded the plain one alone, and
an undecided subgroup or a root group keep what they had. openWallet will
not build a plain statement under a guarded policy, and OpenWalletAction
shows one button with a line saying the decision was the parent's.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
SubgroupManager.refuseKeyState: a child signer refuses a key state whose
birth certificate decides a wallet policy other than the one its own
ceremony announced at the join gate -- undecided included. A ceremony
from an older build announced none and holds the certificate to nothing.
proposeSubgroupKeyState refuses the same thing before asking.
GroupKeyStateManager.parentageOf drops a state whose certificate's policy
does not read or contradicts its guardian keys, so a corrupt policy never
reads as undecided.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
proposeBirthCertificate reads the policy off the child's ceremony: guarded
carries the grant and `guarded`, unguarded carries no keys and
`unguarded`, and an older ceremony with none is certified as it always
was -- or as guarded under a guarded parent. An unguarded ceremony under
a guarded parent is refused with the remedy, start again.
refuseGrant becomes refuseCertificate. Every parent signer refuses a
certificate whose policy does not read or contradicts its keys, and a
guarded group's signers refuse anything but a guarded one, undecided
included -- D4, read off the signer's own key state. ProposedEvent says the
decision in both directions.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
SelectSubgroupAdminsScreen gains a wallet section under the quorum
stepper: guarded (the default) or unguarded, carried to proposeRitual.
Under a guarded parent the unguarded row is shown disabled with its
reason, setWalletPolicy will not move off guarded, a retry is pulled back
to guarded, and refusal() refuses it -- D4. A parent whose guardian keys
do not read cannot make a guarded subgroup, which is now refused before a
ceremony rather than after.
SubgroupManager.walletPolicyOf is the one reader of a group's own policy,
reached through ChatRepository with guardiansOf. DkgJoinApprovalScreen
tells an admin what the subgroup's wallet will be before they join.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
proposeRitual takes a walletPolicy and puts it on the proposal as a
["wallet", ...] tag, beside the parent and subject; acceptProposal files it
on DkgSession.walletPolicy. A proposal whose policy does not read, or that
carries one with no parent, is dropped, and proposeRitual refuses a policy
on a ceremony with no parent before anything else.
Schema 28 -> 29: the nullable DkgSession.walletPolicy, an auto-migration
like the other nullable additions.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
SubgroupWalletPolicy and WalletPolicyTag, ["wallet", "guarded"|"unguarded"],
and SubgroupBirthCertificateEvent.parseWalletPolicy. A certificate with no
tag reads Undecided; one whose tag does not read, or contradicts the
guardian keys beside it, reads null rather than undecided. certifies is
unchanged. Nothing writes the tag yet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
FrostSigningEvents.DECLINE and MarmotDirectMessage.ENVELOPE_KIND are both
30330. Recognising protocol kinds first (ceremony-creation Phase 6) returned
before the direct-message branch, so every direct message in a
current-profile room was dropped unopened. A 30330 that holds a gift wrap is
an envelope; only one that does not is a decline.
ProtocolKindDispatchJvmTest now opens an envelope through
fromGroupEventResult, which no test did before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GroupWalletRollout.GUARDIANS_ON_MAINNET is on, lifted by decision ahead of
the signet run its condition named. The mixed-version risk stays open and
is said so in the gate and the guardians note; guardian spends on mainnet
are held to MAINNET_CAP like every other spend.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A subgroup's ceremony runs in its parent's room, so that room holds a
certificate and a key-state signing session for every subgroup made under
it. The ceremony screen took the newest of each kind regardless of which
subgroup it was for, so a parent's second subgroup opened on the first
one's finished certificate, with "Ask the parent group to certify"
already disabled.
Each session is now matched on what its event names: the certificate on
the subgroup's threshold key, the key state on the ceremony. The newest
match still wins, so a failed request can be asked for again.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Found on the emulators: refusing a Marmot room left the app bar reading
"Shared key for " and nothing. It reads "Shared key" until the room is
loaded. The plan now records what was checked on the three emulators that
found the bug, and what was not.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Schema 26 -> 27. MIGRATION_26_27 deletes the chat lines the removed document
arms wrote and the unsupported bubbles a subgroup's ceremony left in its
parent's chat -- found by the kind of the inner event behind them, so an
unknown payload's bubble stays -- then drops the twelve document tables,
children before parents, and MarmotInnerEvent.payloadEventId. Manual,
because it deletes rows. GroupSignedEvent keeps whatever the group signed,
and the chronicle keeps its column and its lines.
The nip30303 events and tags, the Mantra entities and DAOs, the repository
interface and the Markdown helper go with the tables, and so do the tests
that only exercised them. The key-state test signs posts where it signed
dialects. The migration test builds v26 from the exported schema and checks
the result against v27's.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The inbound path no longer opens document submissions or applies document
kinds, and ProposedEvent no longer describes them. Nothing could send one:
the screens that did were removed in part 1.
The member chronicle carried nothing but document kinds. It is kept, and
what it may carry is now a ChronicleKinds value -- empty in the app, so a
device asks for no chronicle, sends none, and applies nothing from a page an
older build sends. The row-rebuild fallback and translation supersession go
with the documents. The chronicle's tests exercise the machinery with a kind
set of their own; the round trip through document rows is deleted with the
rows, as are the document-only apply test and the tests of translation
supersession.
Tests that used a dialect as an arbitrary signed payload now use a post.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Nothing navigated to the artifact, chapter, dialect or translation screens:
ArtifactDetailRoute, AddArtifactRoute and AddDialectRoute were never built,
and the other six were only reached from inside those screens. Removed with
their view models, UI states, routes, nav host entries, the repository behind
them, TranslationScaffold, DetailRow, and the 65 strings only they used. The
group details screen loses the document repository it stored and never read.
The string catalogue test now checks its escapes on a string that stays.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A subgroup's ceremony runs in its parent's Marmot room, where its kinds
30310-30312 are also nip30303 document kinds. Round 1 (30312) was read as a
document submission first, failed to parse, and reached every other admin's
chat as an unsupported bubble of raw hex; the proposal and host keys reached
the translation arms before the ceremony's own. fromGroupEventResult now
stores a protocol message and returns before any document handling, and
applyInnerEvent checks the protocol arms first.
The comments that called the overlap safe are corrected, and the
"No app logic to handle" line every application message logged at info is
now a debug line that does not read like a failure.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
There was no way to end a ceremony: ChillDKG has no "no" short of it, and
the approval screens pointed at an abandon action on the ritual screen that
did not exist. ChillDkgRitualManager.abandon now ends a running ceremony this
device is in, filing and broadcasting the same FAILURE a fault sends, and the
ceremony screen offers it behind a confirmation.
record() now ignores a FAILURE from anyone outside the ceremony. In a
parent's room every member can read a subgroup's proposal, so any of them
could otherwise end it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
DkgRitualRoute and the three approval routes now require the ceremony's id,
and the screen loses its start button, its quorum stepper and its fallback to
the room's newest ceremony. A transcript line naming no ceremony opens
nothing. The title is the ceremony's name rather than the room's, so a
subgroup's ceremony is not called after its parent.
Try again on a failed ceremony goes back to the screen that starts that kind:
the creation screen for a group's own key, or the subgroup picker filled in
from the ceremony that failed. Retrying a failed subgroup used to open a
whole-room ceremony in the parent's room.
A test counts the screens that open a ceremony, so a fourth is a decision.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
CreateSharedKeyScreen takes the quorum and a single button, and nothing
else. It refuses a Marmot room and a group of one with a reason, hands over
to a ceremony the group already has rather than offering a second, and
replaces itself with the ceremony once it is opened so back never returns to
an offer. The group details' Shared key button and the card shown when a
robust group was created without its ceremony both lead here.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Three ways the shared-key screen could lose the ceremony it was showing, and
so offer to start one:
- initiate() ran again whenever the screen came back into view, resetting
the state to no ceremony and stacking another set of watchers. It now runs
once per view model.
- five coroutines each read the state, changed one field and wrote it back,
one of them across a suspending query, so an old copy could overwrite a
newer one. The state is now built by one flow and written in one place.
- an approval closed its screen as the button was pressed, cancelling the
view model's scope and with it a publish still in flight. The screen now
closes once the approval has gone out.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
proposeRitual now refuses a ceremony with no parent in a Marmot room, and
acceptProposal drops one arriving there. A Marmot room's id is derived from
the key it already has, so a key of its own is one it could never use -- and
the shared-key screen offering to start one in a parent's room is how a
second, whole-room ceremony opened beside a subgroup's.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A guarded wallet's coins have a spender outside the room: a guardian can
take a coin the child's open spend wants, and the child can move one a
guardian's wants. Neither winner is a row in the loser's room, so the loser
sat signed and rebroadcast forever. The watcher now asks the chain, for a
signed spend Electrum has not heard of, which transaction took each coin;
once that one confirms, recordReplaced marks the spend replaced and settles
every other spend of the coin, as recordConfirmed does for the room's own.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Whenever a guarded wallet's coins change, the wallet screen reads every used
address's history off Electrum, recognises a guardian's leaf in the witness
that spent a coin, and writes one transcript line per transaction saying
which group above took how much -- the child's only notice, since nobody in
its room signed it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GroupSpend checks a spend as a guardian would sign it: every input under
the descendant's account, locked by the leaf's 12 blocks, signed over the
script-path sighash as the key the ancestor's walk reaches, and witnessed
with the leaf and its control block. The builder waits out the lock and
weighs the leaf; the signing manager proposes, accepts, signs and completes
it, and every signer reads the descendant's wallet from its own copy of the
statement. A guarded wallet row on the ancestor's wallet screen opens the
send form for it. Schema v27 adds GroupTransaction.guardedChatRoomId.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
bitcoin-kmp reads a taproot derivation's fingerprint back from a serialised
PSBT as a signed int, so for any account whose fingerprint has its top bit
set -- half of them -- every member but the proposer refused every spend as
not the wallet's own. Compare the fingerprint as an unsigned 32-bit value.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A member in both rooms carries a descendant's signed guarded statement into
an ancestor's MLS room as a 30333 notice, and can pass one on a generation
further. Every ancestor device believes it only when its own guardian root
reaches the key the statement names for it. The wallet screen lists the
guarded wallets below the group with what is within its reach, and the
statements this device can still carry up.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A room whose certificate grants guardian keys is offered a guarded wallet,
with a plain one beside it, on test networks; mainnet waits behind
GroupWalletRollout.GUARDIANS_ON_MAINNET. A spend's account is the one the
statement opened, so its own spends sign the guarded addresses' key path,
and wallets are watched per descriptor rather than per xpub.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
An account knows its guardians and puts a leaf per guardian in every
address, internal key unchanged. The statement names them and carries the
certificate that granted them, and verifies only when the certificate
certifies this room and grants exactly those keys. A guarded statement
outranks a plain one for its account, and its first guardians stand.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The parent grants the child its own guardian key and every key its own
certificate carries, one step further each, as guardian tags. Every parent
signer recomputes the grant and drops a certificate proposal carrying any
other keys before it is shown; one carrying none still signs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GuardianDerivation: a group's guardian root, the child index, the grant a
certificate carries, the 12-block leaf, the caterpillar tree, the descriptor
fragment and the cache an ancestor signs a leaf with. Pinned vectors checked
against an independent implementation.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The form read which coins and change addresses open spends hold once,
when it opened. A spend another member proposed while it was open still
looked free, so sending everything would ask the group to sign a second
spend of coins only one spend can have. The form now combines the chain
state with observeGroupTransactions, as the wallet screen does, and
re-reads both reservations on every change, as one state.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The group wallet's address list: every paid receive address, the
current one and the two after it, any unused one handed out as a QR
code, and the used change addresses -- docs/group-wallet-addresses.md.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The amount field shows its unit, how many sats are confirmed and free to
spend, and on mainnet the cap; a problem with the amount -- not whole,
too large, more than the group has, too small, over the cap -- is said on
the field in its error state instead of under the form. The send-everything
row is one 48dp toggleable target, and turning it off again finds the
amount typed. The wording of every reason is one function shared by the
field and the preview.
GroupSendScreenJvmTest types into the form on regtest and mainnet.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
SEND_AN_AMOUNT kept every network on send-all until a release had shipped
with no mislabelled change, which no release could do while the gate was
closed. It is removed: the send screen always offers the switch and the
amount, and on mainnet the existing 100,000 sat cap bounds what a change
bug could put at risk. Phase 3's end-to-end test is what stands in for
the release the gate was waiting for.
The view model test no longer branches on the constant, and a mainnet
test holds an amount to the cap.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GroupWalletState.changeAddresses lists the change addresses that have
been used, newest first -- never a current or a next one, since change
is chosen by a spend. The address list shows them in a section of their
own once there are any, with copy and nothing else.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Tapping the current or a next address on the list opens it in a sheet
-- the receive card's QR code, selectable text and copy, titled with the
index so a member can say which one they gave. A used address has copy
and nothing else: copying is for checking, a QR code is for being paid.
The sheet closes itself, with "Address N was just paid", when Electrum
reports a payment to the address it is showing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Two open spends with change both paid to the first unused change
address, since a spend waiting for signatures has not used it yet.
Change indices are now reserved like coins: reservedChangeIndices reads
the claims of every open spend, and the builder takes the lowest watched,
unused, unheld index. It chooses among generated addresses only, so
change never lands outside the window; CHANGE_LOOKAHEAD rises to 10 and
past it a spend with change returns TooManyOpen.
The change branch not having answered is FindingChange rather than
NothingToSpend, and change too small to send back is reported as
Built.absorbed and said in the preview.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GroupWalletAddressesScreen, pushed from an "All addresses" button in the
wallet screen's receive card: index, address, and what each has been
used for, newest first, with the current address on the secondary
container and a copy button on every row. Its view model reads the
wallet GroupOnchainWallets already holds, so the list costs no
subscriptions and agrees with the wallet screen about which address is
current.
FakeElectrum moves out of GroupOnchainWalletJvmTest so the view model's
test can drive a chain too.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GroupWalletState.receiveAddresses: every paid receive address, the
current one, and ADDRESSES_AHEAD (2) after it, newest first -- reaching
the highest paid address even when a watch-only wallet was paid past
that. Null until Electrum has answered, as receiveAddress is.
The test fake now reports a status over a script's history rather than
its unspents, as Electrum does, so an address spent from stays used.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The gated amount field read its text by dropping every non-digit, so
21000.5 became 210,005 sats and 0.0005 became 5. parseSats accepts
thousands grouping and refuses decimals, grouping that groups nothing,
non-ASCII digits and more sats than exist; the preview reports which.
The field shows plain digits grouped and keeps what was typed.
Also plans the rest in docs/group-send-an-amount.md: send-an-amount
opens on every network at once, held on mainnet by the existing cap.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The wallet screen's title, the room's detail row, the button that opens
one and the proposal summary a member agrees to all said "group wallet".
They now say "group onchain wallet", which is what it holds, and agrees
with GroupOnchainWallet in the code. Sentence case, as every string is.
Resource keys are unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
DERIVATION_PAYLOAD closed its JSON object with a bare `}`. The JVM reads
that as a literal; ICU, which backs Regex on Android, rejects the
pattern. It was a private val of FrostSigningManager, so the throw came
from the class initialiser, which Android does not retry: every signing
session on the device failed with NoClassDefFoundError until restart.
The first symptom was a ceremony that "couldn't ask the group to agree
its signing key".
The JVM suite cannot ask ICU, so RegexPortabilityJvmTest reads every
Regex literal out of the main source sets and holds it to the stricter
rule: a `}` is escaped, inside a class, or the end of a quantifier. With
the old pattern restored it fails on FrostSigningManager.kt.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every conflict was the curated-lists removal on bennu meeting a wallet
line added beside it: the removal is taken and the wallet lines kept.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
And its registry, GroupChainWallets, is GroupOnchainWallets, with the
files, the test and the plan's references following.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GroupWalletRollout holds the three gates, each with what lifts it. The
flag is on, its condition met. Send-an-amount is off, so the form sends
everything confirmed to one address and says so. A spend proposed from
this device on mainnet may send at most 100,000 sats; test networks
have no cap. The gates are on the composer and nothing else: a device
behind them still accepts, shows and signs what another member
proposes, through every check GroupSpend makes.
The plan is marked built, with its table of where the build chose
differently. Full JVM suite: 1474 tests, none failing; the Android
target compiles; every M3 budget is met.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Every spend the room proposed, newest first, for every member whether
or not they have a node: what it sends and where it has got to --
waiting for signatures, abandoned, signed, sent (reported back by
Electrum, not merely broadcast), confirmed, or replaced by a later
spend. A signed spend Electrum has not reported can be sent again; a
signed, unconfirmed one can be bumped to twice its feerate, a new
session over the same coins that whichever confirms settles. Each row
opens its signing session.
Full JVM suite: 1473 tests, none failing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
GroupTransactionWatcher, one per identity beside its wallets: every
signed spend still on its way is sent from this device too, sent again
every ten minutes while Electrum has not reported it -- its broadcast
reports success whatever the server said -- and followed to a
confirmation. FrostSigningManager.recordConfirmed then settles every
other spend of the same coins: one still collecting is abandoned as
spent by another transaction, one already signed is marked replaced
rather than failed, and the room gets one line saying it confirmed.
GroupSpendBuilder.bump re-spends the same coins at a higher feerate,
the extra taken from the change or, for a send-all, the one payment.
The test caught a stale write: a pass followed each spend from the list
it started with, so settling one spend was undone by writing the next
back from its old copy. follow() reads its row again first.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>